Understanding the molecular mechanisms underlying graft ... · RESEARCH ARTICLE Open Access Understanding the molecular mechanisms underlying graft success in grapevine M. Assunção1*,

RESEARCH ARTICLE Open Access

Understanding the molecular mechanismsunderlying graft success in grapevineM. Assunção1* , C. Santos2, J. Brazão3, J. E. Eiras-Dias3 and P. Fevereiro1,4

Abstract

Background: Grafting is an intensive commercial practice required to protect the European grapevine against thePhylloxera pest. Rootstocks resistant to this pest are hybrids of American vine species with different levels ofcompatibility with European Vitis vinifera varieties. Aiming to understand what drives grafting compatibility ingrapevine, a transcriptomic approach was used to search for master regulators of graft success. Two scion/rootstock combinations, with different levels of compatibility, were compared in a nursery-grafting context attwo stages, at 21 and 80 days after grafting.

Results: In the most compatible combination, an earlier and higher expression of genes signaling the metabolic andhormonal pathways as well as a reduced expression of genes of the phenolic metabolism and of the oxidative stressresponse was observed. At 80 days after grafting a higher expression of transcription factors regulating vascularmaintenance, differentiation and proliferation was obtained in the most compatible combination. Moreover, lowerexpression levels of microRNAs potentially targeting important transcription factors related to plant development wasobserved in the more compatible combination when compared to the less compatible one.

Conclusion: In this context, a set of regulators was selected as potential expression markers for early prediction of acompatible grafting.

Keywords: Grapevine, Grafting, Graft compatibility, Molecular mechanism of grafting, Vascular differentiation, Transcriptionalregulation of grafting, Post-transcriptional regulation of grafting

BackgroundGrafting is a very ancient method used worldwide forclonal propagation, to reduce juvenility and to overcomemany biotic and abiotic stresses. Nowadays it is a wide-spread technique used in fruit trees, vegetables and flowerproduction and therefore with an enormous agriculturaland economic impact [1]. In the European viticulture,grafting is almost imperative due to phyloxera, an insectthat feeds from the roots of Vitis vinifera cultivars leadingvines to death. So far, the only way to overcome this pesthas been to graft the European cultivars in American orAmerican hybrid resistant rootstocks [2]. Grapevine graft-ing may be subjected to incompatibility since two genetic-ally different entities are put together. Graftincompatibility may be defined as the failure to form a

successful graft union between two plant parts, i.e. the failto form a proper functional composite plant when allother requirements, such as technique, timing, phytosani-tary and environmental conditions are satisfied. Incom-patibility may be expressed even after many years ofnormal growth, not only in grapevine but also in otherspecies such as pear-quince, apricot, and pear [3–5]. Mostof the incompatibility studies were directed to morpho-logical and physiological observations between compatibleand incompatible unions [6, 7]. The development of agraft union starts with the healing of the graft zone by awound response where the living regions of the scion andstock in contact initiate the proliferation of parenchymat-ous cells that origin a mass of undifferentiated cells calledcallus. This callus works as a bridge between the two plantparts until differentiation of the new cambial cells intoxylem and phloem tissues, enabling the vascular connec-tion between scion and rootstock [6, 8].Early detection of graft incompatibility is of great im-

portance for nurseries and farmers since it could be used

© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

* Correspondence: [email protected] Cell Biotechnology Laboratory, Instituto de Tecnologia Química eBiológica António Xavier (Green-it Unit), Universidade Nova de Lisboa,Apartado 127, 2781-901 Oeiras, PortugalFull list of author information is available at the end of the article

Assunção et al. BMC Plant Biology (2019) 19:396 https://doi.org/10.1186/s12870-019-1967-8

http://crossmark.crossref.org/dialog/?doi=10.1186/s12870-019-1967-8&domain=pdfhttp://orcid.org/0000-0003-3143-7427http://creativecommons.org/licenses/by/4.0/http://creativecommons.org/publicdomain/zero/1.0/mailto:[email protected]

as a tool for early selection of best graft combinations.Many detection methods have been developed such as invitro techniques [3], histological observation [9, 10] iso-zyme analyzes [11, 12] and phenolic analyzes [4, 13–15].These are important methods, however these phenologicand metabolic markers vary a lot depending on the graftingcombinations, environmental and soil conditions, makinggraft incompatibility detection very difficult to perform.Only recently the molecular mechanisms associated withgrafting began to be studied. In a broad approach, Cooksonet al. [16] studied the transcriptional profile of graft unionin grapevine autografts. They found genes differentiallyexpressed between 3 and 28 days after grafting to be re-lated to cell wall modification, wounding, hormone signal-ing and secondary metabolism. In a later study, Cooksonand collaborators also compared the gene expression pro-file at the graft union of heterografts (rootstock and scionfrom different genotypes) with autografts (rootstock andscion from the same genotype) and found up-regulation ofgenes involved in stress responses, suggesting that it couldbe related to the detection of a non-self grafting partner[17]. Irisarri et al. [18] detected a higher accumulation ofantioxidant gene transcripts in compatible grafts early ingraft development of pear/quince and suggested to be asso-ciated with better protection of the damaged tissues.Melnyk et al. [19] reported that in Arabidopsis grafts thexylem is formed only after the phloem connection, and forthe phloem connection a group of auxin response factorsact below the graft junction, such as AXR1 and ALF4.Interestingly, mutating AXR1 in the upper side of the graftunion rescued the phloem connection in a rootstockmutant of AXR1 [19]. These recent findings attribute animportant role in the spatial and temporal communicationbetween scion and rootstock to the process of graft con-nection. Chen et al. 2017 analyzed the transcriptome ofLitchi compatible autograft and incompatible heterograft at2 h, 4 days and 21 days after grafting. The results sug-gested that genes expression related to wound response,auxin (IAA) and signal transduction pathways couldhave a key role in Litchi grafting healing process. Morerecently, the important role of auxin was also reportedin Citrus compatibility studies. [20] Although the re-cent findings, the molecular mechanisms of grafting arestill largely unknown, particularly in woody plants.By studying and understanding the molecular mecha-

nisms of the union formation in grapevine we will be closerto develop molecular markers for compatibility, less vari-able and more suitable for breeding. Having that in mindwe compared two heterografts of two clones of the cultivarTouriga Nacional grafted in the rootstock Richter110 whichshowed different levels of compatibility. Touriga Nacionalis presently the main grapevine cultivar used to producewine in Portugal and it is known to produce the best qualityPortuguese wines. Richter 110 (110R), a worldwide used

rootstocks, has shown to have deficient compatibility withSyrah cultivar [21] and reported to have different levels ofgraft compatibility with Touriga Nacional (TN). We ex-plored, in a nursery-grafting context, two different timepoints; 21 days after grafting, when grafts are taken out ofthe callus induce chamber to be transferred to the fieldand, 80 days after grafting, i.e. after 2 months in the fieldand when the root system is already developed. By lookingat the differentially expressed genes and micro RNAs (miR-NA)s between the more and the less compatible combin-ation we aimed not only to unveil what mediates graftcompatibility but also to find master regulators of graftunion formation that could be used as molecular markersfor early prediction of graft success in grapevine.

ResultsAnalysis of grafting successThe graft compatibility of the combinations TN21/110Rand TN112/110R was accessed quantifying the graft suc-cess, i.e. counting the number of grafts with well-developedroot and shoot system and with a well-established union atthe end of the vegetative cycle. Three independent trials,performed in independent years, revealed TN21/110R com-bination was more successful than TN112/110R (Fig. 1).Touriga Nacional grafted on the rootstock 1103-P was alsoevaluated, and TN21 proved again to have a higher graftsuccess (66.5%) when compared with TN112 (52.3%).The mean comparison of the graft success over 3 years

of clones TN21 and TN112 over 110R revealed a p-valueof 0.169. When adding the comparison of the graft suc-cess obtained with 1103-P, the p-value retrieved was of0.051. Based on this, it was considered that there is a dif-ference in graft compatibility in between the two clonesbeing the TN21/110R combination the more compatibleand TN112/110R the less compatible one.To study the molecular mechanisms responsible for

these different compatibility levels, samples were collected21 and 80 days after grafting (DAG). At 21DAG, fewunsuccessful or null grafts where detected, characterizedat this time point by the absence of callus or the presenceof dead tissue in the area of the scion and rootstock incontact. At this stage, there is no incompatibility per se,the formation of the callus is a consequence of the re-sponse to the wound and to the phytohormones appliedin the wax during the graft procedure (see Methods sec-tion), and there is still no interaction of the two vascularsystems. Incompatibility is only detectable after the graftshave been transferred to the field. At 80DAG, the majorityof unsuccessful grafts were observed, i.e. grafts that failedto form a union between two plant parts at graft interface(Additional file 1).Autografts are usually reported to be compatible since

it is expected no incompatibility when a genotype isgrafted onto itself. Autografts of the Touriga Nacional

Assunção et al. BMC Plant Biology (2019) 19:396 Page 2 of 17

clones and of the 110R rootstock were evaluated in the sec-ond and third field trial (Fig. 2). In 2015, the two autograftsof Touriga Nacional had different graft success rates, withthe autograft TN112/TN112 exhibiting a success rate of92% while TN21/TN21 had only 50% of success and the110R/110R only 56%. In 2017 success rates were compar-ably higher than in 2015, with TN112/TN112 showing, asin the previous trial, the highest compatibility rate. Theseresults led to the exclusion of autografts as compatiblecontrols in the transcriptomic approach.

Genes differentially expressed between the twoheterografts at 21DAGThe transcriptome analysis revealed 33 differentiallyexpressed genes (DEG) between the more compatible

combination (TN21/110R) and the less compatible one(TN112/110R) at 21DAG. In Table 1 it is shown theDEG at 21 DAG, where 17 transcripts are up-regulated(more abundant) in the more compatible combination(TN21/110R) and 16 are down-regulated in the morecompatible when compared to the less compatible com-bination (TN112/110R).Transcripts belonging to the BIN categories “Photo-

systhesis”, “Cell wall”, “Tetrapyrrole synthesis”, “Poly-amine metabolism”, “Miscellaneous”, “RNA” and “DNA”and “Signaling” were found to be more expressed in themore compatible combination (TN21/110R) at 21DAG.From the genes more expressed in the more compat-

ible combination it was found an expansin gene (VIT_01s0026g02620), codifying for a protein that responds to

Fig. 1 Percentage of graft success in the heterografts. Graft success was evaluated at the end of the vegetative cycle for Touriga National clonesTN21 and TN112 grafted on the rootstock 110R, in 2012, 2015 and 2017, and for the same clones grafted in 1103-P in 2017

Fig. 2 Percentage of graft success in the autografts. Graft success was evaluated at the end of the vegetative cycle for autografts of the rootstock110R (110R/110R) and the scions TN21 (TN21/TN21), and TN112 (TN112/ TN112), in 2015 and 2017


Table 1 Differentially expressed genes at 21 DAG between the more compatible (TN21/110R) and the less compatible (TN112/110R)combination


auxin signaling, involved in cell wall loosening and cellelongation in acidic growth [22]; an ethylene responsivetranscription factor, (VIT_00s0662g00030), member ofthe AP2/ERF TF superfamily known to be an importantregulator of developmental processes and to responsesof various types of biotic and environmental stresses[23]; and a EP3 chitinase (VIT_05s0094g00330).The category “DNA” comprises a histone H2A.6 (VIT_

00s0868g00020) up-regulated in the more compatibleheterograft and the category “signaling” contains threetranscripts, two differentially more expressed in theTN21/110R heterograft: one protein kinase, one ethyleneresponse factors, Erg-1 (VIT_07s0005g00870), with a BINdescription of signaling in sugar and nutrient physiology.This ethylene response factor has a log2 fold change of5.4, the highest of this analysis. Contrastingly, three genescodifying enzymes of the phenylpropanoids: two stilbenesynthases, (VIT_16s0100g01100, VIT_16s0100g01200) dir-ectly involved in the synthesis of resveratrol and one Or-cinol O-methyltransferase1 (VIT_12s0028g01940), whichare more expressed in the less compatible combination(TN112/110R).

Genes differentially expressed between the twoheterografts at 80DAGThe transcriptome analysis revealed 63 DEGs betweenthe two heterografts at 80DAG. In Table 2, it can beseen the DEG, 26 of which are up-regulated and 37 down-regulated in the more compatible combination (TN21/110R) in comparison to the less compatible one.The more compatible combination (TN21/110R) showed

a higher abundance of transcripts associated with cellularprocesses, like the categories “Photosynthesis”, “MajorCHO” and “Cell wall”. This last category comprises threetranscripts: a pectin methylesterase (VIT_02s0025g04210),a UDP-glucose 4-epimerase (VIT_15s0048g00510) and anexpansin (EXPA15, VIT_01s0026g02620).Regarding the less compatible combination, DEGs be-

longing to categories “RNA”, “Secondary metabolism”,“Minor CHO metabolism”, and “Miscellaneous” werefound to be more expressed in the less compatible com-bination (TN112/110R). The category “RNA” includedthree TFs, a NAC (VIT_15s0048g02270) and two WRKYs,WRKY18 (VIT_04s0008g05760) and WRKY70 (VIT_08s0058g01390), as well as an RNA-dependent RNApolymerase 1 (RDR1, VIT_01s0011g05880) and a DNAmethyltransferase (VIT_12s0034g02560). WRKY TFsplay a dual role in the brassinosteroids-mediated regu-lation of PAMP (pathogen-associated molecular pat-terns)-triggered immunity (PTI) signaling [24]. In thecategory “secondary metabolism”, two stilbenesynthases (VIT_16s0100g01190, VIT_16s0100g00780)were also more expressed in the less compatible com-bination (TN112/110R). A resveratrol synthase (VIT_

16s0100g01110) was also more expressed in the less com-patible combination. Stilbene synthases are precursors ofresveratrol, which have been related to the initiation ofthe hypersensitive response in Vitis, and related to pro-grammed cell death (PCD) [25].Within the category “Miscellaneous” two transcripts

highly expressed in the less compatible combinationannotated as Arachidonic acid-induced DEA1 (VIT_12s0055g00800 and VIT_02s0154g00280) were found.Still in this category, transcripts coding antioxidantproteins were found all more abundant in the less com-patible combination, like two short chain dehydrogen-ase, SDR (VIT_14s0068g01760, VIT_08s0040g01200), aglutathione S-transferase (VIT_08s0040g01200), a per-oxidase (VIT_18s0001g06850), and two lipid transferprotein (VIT_02s0154g00280, VIT_12s0055g00800).

Transcriptome profiling between 21DAG and 80DAGTranscriptome analysis was also performed betweentime points (from 21DAG to 80DAG) for each com-bination. A total of 697 genes were commonly DE inboth heterografts (Fig. 3), whereas 411 DEGs wereonly DE in the more compatible combination (TN21/110R) and 416 only in the less compatible combin-ation (TN112/110R) (Fig. 3a).The functional categories enrichment of the common

DEG from 21 to 80DAG (Fig. 3 b) showed an over-repre-sentation of the up-regulation on the transcripts “Photo-synthesis”, “Lipid metabolism”, the subcategory “signaling.receptor kinases”, “amino acid metabolism”, “secondarymetabolism”, the subcategory “Stress.abiotic.heat”, “Trans-port” and Miscellaneous subcategory “misc.gluco-,galactoand manosidases”. Oppositely there is an over-representa-tion of the down-regulation of the categories “RNA”, theMiscellaneous subcategory “misc.cytochrome P450” as wellas the “not assigned” category.The categories “PS”, “minor CHO metabolism”, “lipid

metabolism”, “stress. Abiotic”, “protein” and “transport”were enrich from 21 to 80DAG in the less compatibleTN112/110R combination (Fig. 3d). In the more com-patible combination (TN21/110R) there is an enrich-ment of down-regulated transcripts of the categories“Cell Wall”, “hormone metabolism”, “RNA”, “signaling”and “Cell” from 21 to 80DAG (Fig. 3c).Due to the importance of these categories in the graft

healing process and in the vascular differentiation, thesetranscripts are presented as a heatmap for both combi-nations(Fig. 4). For this, some transcripts annotated byGrimplet et al. [26], as belonging to the categories underanalysis, were rescued from the data sets and added tothose obtained by the BIN categorization. The clusteringof this group of transcripts, using an average linkageclustering method with Kendall’s Tau distance measure-ment, revealed the difference in the expression timing


Table 2 Differentially expressed genes at 80DAG between the the more compatible (TN21/110R) and the less compatible (TN112/110R) combination


between combinations, where a higher number of thesegenes was more expressed at 21DAG in the more compat-ible combination. When clustering by categories and subcat-egories the same group of genes (Additional file 2), andlooking at the expression in all libraries it is possible tosee, despite the same profile of up/down regulation, thedifferences in the expression level. In the specific DEGof the more compatible combination (TN21/110R),transcripts of the auxin and ethylene signaling pathwaysare more expressed at 21DAG. Of the 10 transcripts in-volved in the auxin signaling pathway, seven were moreabundant at 21DAG and only three are up-regulated at80DAG. For the ethylene signaling pathway, the samepattern is observed, seven transcripts being more abun-dant at 21DAG, and only three up-regulated at 80DAG.In contrast, there was a higher expression at 80DAG ofmost of the transcripts from the signaling pathway ofthese two phytohormones in the less compatible com-bination (TN112/110R). Specifically, DEG in this combin-ation, 10 out of 13 transcripts of the auxin signalingpathway are more expressed at 80DAG, while from fourethylene signaling related transcripts, three are up-regu-lated 80DAG. Furthermore, more transcripts related tothe stress hormone abscisic acid (ABA) are differentially

expressed in the less compatible combination, being sixout of nine genes overexpressed in the less compatiblecombination at 80DAG.In total, 138 TFs were differentially expressed from 21

to 80DAG in both combinations, 57 commonly present inboth combinations, 42 exclusively in the TN21/110R and39 exclusively in the T112/110R combination. Some ofthese TFs are directly involved in the maintenance anddifferentiation of vascular cambium cells, such as VIT_19s0027g01120 (LBD4), VIT_18s0001g10160 (WOX4)and VIT_18s0001g06430 (ATHB-6) [27–29].TFs regulating auxin and ethylene pathways were

also differentially expressed between time points, aswell as, diverse MYB, WRKY, and C2H2 zinc fingerTFs family members.

Differential expression of transcription factors involved inthe regulation of vascular differentiationDue to the importance of the vascular tissue connectionin the development of a successful graft, detailed expres-sion quantification was undertaken between the two com-binations and between the two time points on knownregulators (TFs) of vascular differentiation (Table 3).

Fig. 3 DEGs between 21DAG and 80DAG found in both heterografts. DEGs with significant differences were selected based on a FDR < 0.05, alog2 fold change > 1 and more than 100 counts. Venn Diagram with the number of DEGs found between 21DAG to 80DAG in the morecompatible heterograft (TN 21/110R) and in the less compatible one (TN 112/110R) (a). PageMan visualization of MapMan functional categoriesenriched in the DEG found commonly DE (TN21/110R and TN21/110R) (b), specifically DE in the more compatible (TN21/110R) (c) and specificallyDE in the less compatible (TN112/110R) (d). The over representation/under representation of the functional categories in the up/down-regulatedgenes is given by shades of blue and red, respectively


The expression of these eight TFs, previously found inthe transcriptome data with significant differences in be-tween 21 and 80DAG in both TN21/110R and TN112/110R but with no significant differences between them,was quantified by dPCR (Fig. 5).As observed in MACE-Seq, these TFs showed an over-

all down-regulation from 21 to 80DAG, and a good cor-relation (r2 = 0.68) between MACE-Seq and dPCRresults was obtained (Additional file 3).The expression of three TFs, VviLBD4, VviHB6, and

VviERF3 revealed to be significantly different in betweenthe two heterografts. At 21DAG only VviLBD4 is signifi-cantly different between heterografts being down-regu-lated in the more compatible combination whencompared to the less compatible one. At 80DAG,

VviLBD4 is more expressed in the more compatiblecombination, as well as VviHB6 and VviERF3. TheseTFs are involved in the maintenance of cambium activ-ity, growth, and differentiation. These results showedthat the expression of these three TFs is significantly lessreduced from 21DAG to 80DAG in the more compatiblecombination than in the less compatible combination.

Expression of post-transcriptional regulators at the graftinterfacemicroRNA libraries obtained from the total RNA ex-tracted at the graft union tissue were analyzed, andeight miRNAs were selected for expression analysis,based on their abundance in the graft union tissue andin the predicted function of their potential targets

Fig. 4 Heatmap of transcripts involved in hormone signaling, pathway signaling, TF and cell wall functions found differentially expressedbetween time points (21DAG and 80DAG) for each combination. A total of 108 DEGs in TN21/110R and a total of 90 DEG in TN112/110R wereclustered using Kendall’s Tau distance and an average linkage


(Additional file 4). Quantification by qPCR was performedin the samples collected from the graft zone at 80DAG,since it was the time point when the greatest differences ingene expression between combinations were observed. Re-sults are presented in Fig. 6.A lower abundance of the majority of the quantified

miRNAs in the more compatible heterograft was observed

when compared to the less compatible combination. AllmiRNAs analyzed showed significant differences in the ex-pression between heterografts, except for Vvi-mir482.Vvi-miRNA159 and Vvi-miRNA166 target genes involvedin signaling and gene expression regulation. Particularly,Vvi-miRNA166 targets and negatively regulates the ex-pression of TFs of class III Homeodomain leucine zipper

Table 3 Gene ID, functional annotation, description and function of the eight transcription factors analyzed by dPCR

EmsembleGenomes ID

Annotation Predicted (NCBIrelease 101)

Description Function References

VIT_19s0027g01120

Lateral organ boundariesprotein 4 (LBD4)

LOB family protein Cell proliferation in the cambium. Activation of phloemdifferentiation.

Yordanov et al.[28];Guerriero et al.[29].

VIT_18s0001g10160

Wuschel homeobox 4 (WOX4) Homeoboxtranscriptionfactor

Stem cell maintenance in cambium and differentiationand/or maintenance of the vascular cambium.

Hirakawa et al.[30];Suer et al. [31];Guerriero et al.[29].

VIT_18s0001g06430

Homeobox-leucine zipperprotein ATHB-6 (ATHB6)

b-ZIPtranscriptionalfactor

Negative regulator of Abscisic acid signaling (ABA) pathway.Cell division and/or differentiation in developing organs.

Sӧderman et al.[27].

VIT_06s0004g03130

Auxin response factor 4 (ARF4) Auxin responsefactor

Auxin signaling. Organ polarity, vascular development andorgan asymmetry establishment.

Pekker et al. [32];Hunter et al. [33].

VIT_07s0141g00290

Auxin-responsive proteinIAA16-like (IAA16)

Aux/IAA family Auxin signaling. Repressor of ARF response. Plant growth. Korasick et al.[34];Rinaldi et al. [35].

VIT_18s0001g02540

Response regulator ARR9 Type A ARRs Negative regulation of cytokinin signaling. Callus and lateralroot formation.

Perianez-Rodriguez et al.[36].

VIT_04s0008g06000

Ethylene-responsivetranscription factor ERF3

Ethylene responsefactorAP2/ERF

Ethylene signaling. Cell division in developing vasculartissue. Xylem development.

Etchells et al. [37]Vahala et al. [38]

VIT_16s0013g00890

Ethylene-responsive elementbinding factor - ERF1

Ethylene responsefactor

Ethylene signalling. Cell division in developing vasculartissue.

Etchells et al. [37]

Fig. 5 Quantification of TFs levels in the graft union tissue by dPCR. The gene expression level of eight TF was performed by dPCR, at the graft uniontissue in the TN21/110R (more compatible) and TN112/110R (less compatible) heterografts in two time points, 21 days after grafting (DAG) and 80DAG.Error bars represent the confidence interval and different letters represent significant different values (p < 0.05) according to Mann-Whitney test


(HD-ZIP III), such as REV, PHAB, and homeobox 8 like,known to play an essential role on the regulation of thedifferentiation on the vascular system [39, 40]. Vvi-miR-NA159c targets MYB65 and MYB101 and it has been pro-posed to be involved in the promotion of PCD andinhibition of growth by reducing cell proliferation [41]. AllmiRNA predicted targets were checked in the transcrip-tome data, but low abundance or a false discovery rate(FDR) higher than 0.05 was found.

DiscussionTN21 and TN112 have different levels of graftcompatibilityThe influence of the environment in the construction ofa phenotype is observed by the variation of graft successin the 3 years of field trials. Despite this variability ingraft success, the TN21/110R combination was moresuccessful than TN112/110R in three independent years.Furthermore, the grafting of these clones in a differentrootstock (1103-P) showed the same tendency of graftsuccess. Because the incompatibility of grafting can bedefined as failure to form a successful graft union be-tween parts of two plants [7], it is assumed that, in theseconditions, the TN21/110R combination is more com-patible than TN112/110R.When comparing DEGs between autografts and hetero-

grafts, Cookson et al. [17] observed an up-regulation ofplant defense and stress response associated genes at graftinterface. That evidence led those authors to suggest thatcells at the graft interface can detect the presence of selfor non-self-grafting partners. For this reason, in our study,autografts were initially considered as controls for a morecompatible graft union. However, our field trials revealedvariable autograft successes depending on the genotype,

meaning that graft success is genotype dependent even inautografted plants. Only one reference to autograft incom-patibility could be found [42] but only for in vitro auto-grafts of Vicia faba. Hence, the use of autograft controlsmust be carefully considered, as the autograft and the het-erografts mechanism towards the formation of a well-established graft union may be different. For these rea-sons, we excluded autografts as controls for the transcrip-tomic analysis of our assays.The different behaviour of this clones when autografted

or heterografted suggests a rootstock dependent behaviourtowards graft success. Besides, the more compatible be-haviour of TN21 and the less compatible behaviour ofTN112 were both observed in two different rootstocks(110R and 1103-P). It should be noted that both root-stocks are American hybrids of Vitis berlandieri x Vitisrupestris. If the different levels of compatibility observedare specific to this hybrid species remains to be addressed.

Compatibility seems to be related with the activation ofsignaling in an early graft stageThe role of endogenous hormones and other signalingmolecules have been described to be important in regulat-ing the interaction between scion and rootstock towardsgraft union success in Arabidopsis [8, 43, 44]. In grapevine,Cookson et al. [16] associated graft union formation at 3and 28DAG with an up-regulation of genes of cell wall syn-thesis, secondary metabolism, and signaling. In the presentstudy, a greater abundance of transcripts related to thecategory “Signaling” (hormone or pathway signaling) wasobserved earlier (at 21DAG) in the most compatible com-bination when compared to the less compatible one. Par-ticularly, the role of auxin and ethylene in these two timepoints should be further investigated. A higher and more

Fig. 6 Relative expression of miRNAs measured at 80DAG at the graft union. The expression level was detected by RT-qPCR from tissues collected at thegraft union zone of both heterografts. Error bars represent the standard deviation of three biological replications. Asterisk means significant differenceswith p < 0.05 and ‘ns’ non-significant differences, according to one-way ANOVA


precocious abundance of signaling-related transcripts maybe interpreted as a greater potential for communicationbetween the graft and the rootstock at an early stage ofgrafting. This supports the hypothesis that a higher com-munication potential early in graft union might contributeto graft union success.

Genes encoding oxidative stress and wound healingproteins are more expressed in the less compatibleheterograft at 80DAGStilbene biosynthesis is regulated by many different abioticand biotic stresses [45] and wounding is known to activatethe biosynthesis of stilbenes in grapevine berry skin [46],peanut leaves [47] and Scots pine stems [48]. Recently thephenolic analysis of both TN21/110R and TN112/110Rcombinations showed a significant decrease in phenolicantioxidants sooner in the more compatible one [14]. Thehigher abundance of transcripts of stilbene synthase,resveratrol synthase, and transcripts of oxidative stress re-sponse proteins suggests that the less compatible hetero-graft has a higher oxidative environment throughout thegraft process, hampering a good graft union formation.At 80DAG, the less compatible combination presents a

higher number of transcripts involved in the responses tooxidative stress (polyamine oxidase, glutathione S-transfer-ase, galactinol synthase and peroxidases), and in woundresponse. In contrast, the more compatible combinationpresents up-regulated transcripts with important roles inphotosynthesis and in cell growth suggesting that growthand development processes are taking place at 80DAG.At 80DAG the less compatible combination presents a

higher accumulation of NAC and WRKY transcriptionfactors, known to be involved in the regulation of necrosis,immune response, and the salicylic acid pathway [49–51].Furthermore, a transcript encoding a key enzyme in theABA biosynthesis was found more abundant in the lesscompatible combination at this stage. This could be inter-preted as an increase in stress signaling in the less com-patible combination. The importance of the expression ofWRKY TFs in grafting has been also suggested by the ob-servation of a significant down-regulation of VviWRKY18,VviWRKY52, and VviWRKY70 in the shoot apical meri-stem of grapevine heterografts when comparing to auto-grafts [52]. The higher expression of WRKY18 and 70 inthe less compatible combination at 80DAG may suggestthat the healing process of grafting is was not yet over-come at that time point in that combination.Two of the most highly DEGs, more abundant in the

less compatible combination, at 80DAG, are two Arachi-donic acid-induced DEA1. DEA1 has been related to theinduction of PCD in tomato [53]. Cookson et al. [16]found an up-regulation of a DEA1 gene from 3DAG to28DAG and suggested a relation with phospholipid signal-ing processes occurring at the graft interface at this stage

(callus induction). According to Cookson’s observation,the higher abundance of this transcript in the less compat-ible combination at 80DAG (later in graft development)seem to indicate an unfavorable delay in the signaling andin the regulation of PCD processes in the less compatiblecombination. Altogether, these findings suggest a morestressful environment in the less compatible combination,at 80DAG. This may imply that this combination is unableto deal with the stress resulting from the grafting processin an early stage, being this the possible cause of graft fail-ure. A scheme of the gene expression events that lead to acompatible graft is proposed in Fig. 7.

A higher expression of TF involved in cambiummaintenance and vascular differentiation seems to beimportant in driving graft success at 80DAGThe importance of a vascular tissue connection to forman effective graft union has been referred by differentauthors [8, 54–56], although the molecular mechanismand the master regulators involved are still largely un-known. In this study, VviLBD4, a TF involved in the cam-bium maintenance, was differentially expressed betweenthe more and the less compatible combination at 21 and80DAG. LBD4 is a member of the Organ boundary (LOB)family and has been associated to cell proliferation, mainlyin phloem development, in regulating anatomic featureslike the multiseriate rays [28] and to callus maintenance inArabidopsis [57]. Cookson et al. [16] verified that this geneis overexpressed 3DAG and that its expression decreases28DAG. They propose that in the first stage of graft devel-opment, the abundance of this TF could be related to theformation and maintenance of non-differentiated calluscells; and later (at 28DAG), the down-regulation could berelated to the formation of a functional graft union [16].In this study, from 21 to 80DAG an inversion in theexpression levels occurred for VviLBD4 between the moreand less compatible combination. The significant overex-pression of VviLBD4 at 80DAG in the most compatiblecombination, when compared to the less compatible com-bination, suggests the need for maintenance of the expres-sion of this gene to achieve a more compatible graft.From 21 to 80DAG the maintenance of the ethylene

response factor ERF3 abundance in the more compatibleheterograft (TN21/110R) was also observed, while a sig-nificant decrease was observed in the less compatibleheterograft. ERF family members are involved in growth,development, and biotic and abiotic stress responses [58,59]. Brackmann and Greb [58] suggest that the ERFtranscription factors promote vascular cell divisionsdownstream of PXY and WOX4. WOX4 is a conservedTF described to have a significant role in promoting dif-ferentiation and/or maintenance of the vascular procam-bium, the initial cells of the developing vasculature [59].The significant decrease over time in the abundance of


this TF in the less compatible combination may suggesta higher difficulty to maintain the production of cam-bium cells in this combination.The HB6 TF belongs to HD ZIP I family and negatively

regulates responses to ABA [60] ATHB6 was also impli-cated in cell division and differentiation in developing or-gans [27]. Although no difference in the expression ofVviHB6 at 21DAG was detected between the two hetero-grafts, at 80DAG the higher expression in the less compat-ible combination may indicate a lower control over ABAsignaling and consequently an increase in growth inhib-ition. In sum, the maintenance along time of adequate ex-pression of these TFs seems to be crucial for a compatiblegraft union. Based on the expression variation of these TFand their function described in literature, a workingscheme of the transcriptional regulation of vascular differ-entiation in the mediation of compatible grafts is presentedin Fig. 8. The role of these TFs in graft union is furthersupported by Melnyk et al. [61], where these four TFs are

associated to graft by being over expressed in Arabidopsisgrafts when compared to non-grafts and non-graft cuts.Concerning the post-transcriptional regulation of vas-

cular differentiation, except for Vvi-mir482, all the testedmiRNAs are more expressed in the less compatiblecombination at 80DAG. From the tested microRNAs,Vvi-mir159 and Vvi-mir166 target TFs associated withvascular tissue differentiation. Mir159 targets GAMYBTFs that represses MYB33 and MYB65 in vegetativetissues. Deregulation of these TFs inhibits growth byreducing cell proliferation. More recently, these TFshave also been related to the regulation of vegetative de-velopment [41]. Vvi-miRNA166 targets TFs of the classIII Homeodomain leucine zipper (HD-ZIP III) known toplay an essential role on the regulation of the differenti-ation of the vascular tissues [39, 40]. Despite the regula-tory function of these potential targets, no statisticallysignificant variations (FDR < 0.05) were observed in theRNAseq data, when looking at the expression in themore and the less compatible combination libraries at80DAG (Additional file 4). This was also observed in allthe predicted targets of the eight microRNAS studied.One possible explanation is the very low abundanceof the potential miRNAtargets, as the expression ofthe 58 TF within the potential targets of these eightmicroRNAs has a median value in the more compat-ible combination of 59.42 normalized counts, whilethe less compatible has a value of 57.57 normalizedcounts (Additional file 4).

ConclusionWe aimed to understand what drives grapevine graftcompatibility in order to find master regulators of graftunion. The ultimate goal is to use those molecules asuseful markers for early prediction of graft success. Forthat, we compared heterografts from two clones of thesame variety, both grafted to commercial rootstock 110Rin a nursery environment, at phenotypic, transcriptionaland post-transcriptional levels.

Fig. 7 Hypothetical scheme of transcription regulation associated with a compatible graft union. Up arrows indicate more expression, downarrows indicate less expression. Orange arrows refer to the expression at 21DAG and green arrows to the expression at 80DAG

Fig. 8 Hypothetical scheme of the transcriptional regulation ofvascular differentiation in the mediation of compatible grafts.Putative role of VviLBD4, VviERF3, VviWOX4 and VviHB6 in thevascular differentiation of a more compatible graft when comparedto a less compatible one. Arrows indicate significantly different up-regulation in the more compatible heterograft when compared tothe less compatible one, at 80DAG


Graft compatibility seems to be driven by increased ex-pression of signaling genes of the metabolic and hormonalpathways, which occurs in the more compatible combin-ation earlier than the less compatible combination, in thecallus formation phase. This can be interpreted as havingthe more compatible combination a greater potential ofcommunication between scion and rootstock. Also, as aresult of less oxidative stress, the more compatible com-bination shows reduced expression of genes of the phenolmetabolism and oxidative stress responses.The less compatible combination has a delay and a

higher rate of expression of genes related to wound re-sponse and oxidative stress which shows that it deals inef-ficiently with the establishment of a suitable interaction.The 80DAG stage seems to be the more adequate stage

to evaluate compatibility. In fact, the higher expression ofTFs related to cambium maintenance and vascular tissuesdifferentiation and proliferation, as well as, the lower levelsof post-transcriptional regulation of TFs also involved inthe same processes, clearly characterize the more compat-ible combination. In this context, VviLBD4, VviERF3,VviHB6 and Vvi-mir159, 160 and 166 could be used, at the80DAG, as expression markers of a compatible grafting.

MethodsPlant materialTwo registered clones of the Portuguese cultivar TourigaNacional (TN), the clone 21 ISA-PT (TN21) of PORVIDand the clone 112 JBP-PT (TN112) of JBP/Plansel(https://www.vinetowinecircle.com/castas_post/touriga-nacional/), were grafted on the rootstock 110R (110Richter - V. berlandieri x V. ruprestis) and on the root-stock 1103-P (1103 Paulsen – V. berlandieri x V. rupes-tris). These grafts resulted in TN21/110R, TN112/110R,TN21/1103-P and TN112/1103-P heterografts. Autograftsof the TN21, TN112, and 110R were used resulting inTN21/TN21, TN112/TN112 and 110R/110R.All grafts and field trials were performed at Plansel nur-

sery located in Montemor-o-Novo, Portugal (38°39′N and8°13′W). In 2012 trial, 300 TN heterografts in 110R root-stock were established to collect samples for gene expres-sion analysis and to assess the graft successful rate of thetwo clones of TN. To evaluate the graft success rate of theheterografts in different years and to add the above men-tion autograft controls, the trial was repeated with 200grafts in 2015 and 2017. Furthermore, to confirm thebehavior of the TN clones in a different rootstock, in the2017 trial, 100 grafts between the two clones of TN andthe rootstock 1103-P were also performed.

Grafting procedureAll grafting procedures were executed at the Plansel Nur-sery, as normally performed for commercial purposes.Briefly, hardwood cuttings of the plant material

mentioned above were collected in the winter and pre-served at 4 °C until grafting. Just before grafting, one-budcuttings were made for scions and 35 cm cuttings, withnodes disbudded, were used as rootstock. Scions and root-stocks pairs were bench grafted using the omega grafttechnique. The grafts were dipped in paraffin, suppliedwith 8-quinolinol (0.11%) and 2,5-dichlorobenzoic acid(0.004%), at 75–80 °C and put in boxes containing peatand transferred to a chamber at approximately 30 °C and80–90% humidity, for callus induction. After 21DAGcombinations were transferred to the field. For each graft-ing combination three groups of three plants (nine repli-cates) were harvested in two different time points, 21DAGbefore being transferred to the field, and 80DAG in fieldconditions. Only dead plants were excluded from thepools. The pools were made using random but aliveplants, i.e. plants with various levels of development. Sam-ples were stored in a refrigerator at − 80 °C until use. Fromall plants harvested, a longitudinal cut with approximately1 cm length was made at the graft zone. After removingthe bark and the cortex, the remaining tissues (xylem,phloem, and cambium) were scratched and ground in amortar with liquid nitrogen until a thin powder was ob-tained and stored at − 80 °C until RNA extraction.

RNA extraction and purificationTotal RNA was extracted using an adaptation of themethod described by Chang et al. [62] using 0.1 volumesof NaOAc (3M, pH 5.2) and two volumes of ethanol inthe precipitation steps, to increase precipitation of themicroRNAs. For each plant, an independent RNA ex-traction was performed.Total RNA concentration and purity were verified using

a NanoDrop spectrophotometer ND-2000C (Thermo Sci-entific) by measuring absorbance at 260/280 and 260/230and the integrity by electrophoresis on 2% agarose gel (0.5xTBE, Syber safe, Invitrogen). All samples were treated withTurbo DNA-free™ Kit (Ambion, Life Technologies, Ltd.)according to the manufacturer’s instructions. The removalof genomic DNA was screened by PCR, using primers forthe intronic region of the gene VIT_18s0001g10160(Forward-ATAACCTCTCACCACCCAATC and ReverseCTCCAAGATCCCAATCTGTTC). The PCR mixture(final volume of 20 μL) included: 300 ng of total RNA; 3pmol of each primer; 1x PCR buffer (5X Green GoTaq® Re-action Buffer); 2.5mM of MgCl2, 2.5mM of dNTPs mix;and 0.5 U of GoTaq® DNA Polymerase. The amplificationconditions were the following: an initial denaturing step at94 °C for 2min, followed by 30 cycles of 1min at 94 °C, 1min at 55 °C and 2min at 72 °C, with a final extension at72 °C for 2min. The amplified gene products were visual-ized after electrophoresis on a 1% agarose gel (0.5x TBE,Syber safe, Invitrogen). From the nine independent extrac-tions for each experimental condition, total RNA from


https://www.vinetowinecircle.com/castas_post/touriga-nacional/https://www.vinetowinecircle.com/castas_post/touriga-nacional/

three samples was pooled with equal quantities, enablingto have three pooled biological replicates.

mRNA sequencing and expression profilingPools of the total RNA extracted from the heterografts,at the two time points sampled, were sent in biologicaltriplicate for deep sequencing by Massive Analysis of 3′-cDNA Ends (MACE) by GenXPro GmbH (FrankfurtMain, Germany). Twelve different libraries were ob-tained, three biological replicates per each experimental,TN21/110R_21DAG; TN21/110R_80DAG; TN112/110R_21DAG; TN112/110R_80DAG. Libraries wereconstructed and analyzed by GenXPro GmbH as de-scribed by Zawada et al. (2014). The raw sequencingdata of the 12 libraries were deposit in NCBI under theproject PRJNA517111 and it is available in this addresshttp://www.ncbi.nlm.nih.gov/bioproject/517111. True-Quant technology was used to remove duplicate readsfrom the raw dataset, low-quality sequence bases weretrimmed and the poly-(A)-tail was clipped off. The bar-coded samples were sequenced in an IlluminaHiseq2000, with 1 × 100 bps. MACE reads were mappedonto the Vitis vinifera 12X genome assembly. The num-ber of transcripts per gene was normalized by the librarysize of mapped reads multiplied by 1 million. Theresulting contigs of the assembly were annotated byBLASTX to the Swiss-Prot database, CRIBI V1 annota-tion. The normalization and the analysis of the differen-tially expressed genes (DEG) were done with DEGSeqR/Bioconductor package [63]. The analysis was per-formed between heterografts at each time point and be-tween time-points (from 21DAG to 80DAG), for eachheterograft separately. Genes were considered signifi-cantly differentially expressed with an FDR value < 0.05and a log2 fold change threshold of x ≥ |1|. Only tran-scripts with more than 100 counts at least in one of thecondition were considered. Functional characterizationwas performed using the MapMan web tool Mercator(http://www.plabipd.de/portal/mercator-sequence-anno-tation). Protein sequences of all DEG were obtainedusing BioMart in Phytozome v.12 (https://phytozome.jgi.doe.gov/) to create a mapping file for Mercator. Func-tional annotation obtained was crossed with the one pub-lished in Grimplet et al. [26]. An additional excel filecontaining the DEG between heterografts and betweentimepoints is provided (Additional file 5). The functionalcategories of the DEG from 21 to 80DAG were tested forsignificance using the PageMan enrichment analysis ap-plying the Fisher test [64].Heatmaps were constructed using Heatmapper online

tool (http://www.heatmapper.ca/expression/). Clusteringof the DEG between time-points was performed based onKendall’s Tau distance and average linkage, with normal-ized counts of the single libraries of the genes DE.

microRNA sequencingOne pooled sample per condition was sent for miRNAssequencing on an Illumina Hiseq2000 run by Fasteris(Geneve, Switzerland).Polyacrylamide gel was used to select and purify small

RNAs from 18 to 30 nucleotids (nt) followed by thebound of 3p and 5p adapters. Subsequently, cDNA wassynthesized and amplified generating the libraries forIllumina sequencing. Low-quality reads (FASTq value <13) were removed and adaptor sequences were trimmed,using the Genome Analyzer Pipeline (Fasteris) as de-scribed in Galli et al. [65] Sequences shorter than 18 ntand longer than 25 nt were excluded from further ana-lysis. To identify phylogenetically conserved miRNAs,sRNA sequences were mapped to a set of all conservednon-redundant Viridiplantae obtained from miRBasedatabase (Release 19, August 2012) using Bowtie v0.12.7. The frequency of identified miRNAs was ob-tained by aligning the conserved precursors identifiedin this study and the sRNA library using Bowtie v0.12.7 with the default parameters. For target predic-tion, psRNA target database (http://plantgrn.noble.org/psRNATarget/) and literature review were used.

qPCRThree independent samples of each experimental condi-tion were used for cDNA synthesis according to the man-ufacturer’s instructions of qScript™ microRNA cDNASynthesis Kit (Quanta, Bioscience). Briefly, miRNAs werefirst polyadenylated in a poly(A) polymerase reaction andthen qScript Reverse Transcriptase was used to convertthe poly(A) tailed miRNAs into cDNA using an oligo-dTadapter primer. The qPCR amplification reactions wereperformed using the PerfeCTa Universal PCR Primer(specific to the unique sequence of the oligo-dT adapterprimer) and the PerfeCTa® SYBR® Green SuperMix foriQ™, as forward primers the exact mature miRNA se-quences were used (Additional file 6). The reactionconditions were performed according to the qScript™microRNA Quantification System, with a final volumeof 20 μL, 5 ng of total RNA equivalent was used in eachreaction and the 3-step cycling protocol was used toimprove specificity. Reactions were performed in aniQTM 5 Real-Time PCR Detection System (BioRad,Munich, Germany), and melting curve analysis wasperformed to ensure the specificity of primers. Primers effi-ciency was determined using the LinRegPCR program [66].The expression levels of miRNAs were normalized to

small interfering RNA 4 (siRNA4) and siRNA41 usingthe Pfaffl eq. (1 + Efficency)−ΔΔCt method and the auto-graft 110R as a control sample (Pfaffl, 2001). The tworeference genes selected, were pre-screened for their ex-pression and the expression of 5.8 s rRNA, were used asa reference gene in Vitis miRNA expression analysis in


http://www.ncbi.nlm.nih.gov/bioproject/517111http://www.plabipd.de/portal/mercator-sequence-annotationhttp://www.plabipd.de/portal/mercator-sequence-annotationhttps://phytozome.jgi.doe.gov/https://phytozome.jgi.doe.gov/http://www.heatmapper.ca/expression/http://plantgrn.noble.org/psRNATarget/http://plantgrn.noble.org/psRNATarget/

Wang et al. [67]. After analysing their variance be-tween samples using geNorm and NormFinder [68] inthe Genex software (MultiD, Goteborg, Sweden) thesiRNA4 and siRNA41 of Medicago truncatula recentlydiscovered and published by Formey et al. [69], re-vealed to be the best reference genes. Statistical analysiswas performed with Statistica software (Statsoft Inc.,Tulsa, USA). To evaluate the variance between combi-nations, one-way analysis of variance (ANOVA) wasperformed (p < 0.05).

Digital PCRQuantStudio™ 3D digital PCR (QS3D) System (Life Tech-nologies) was used for the absolute quantification of TFsexpression. Three biological samples at 21 and 80DAGcDNA was synthesized using 200 ng of total RNA previ-ously treated by RNase-free DNase and then reversetranscribed using ImProm-II™ Reverse TranscriptionSystem following the manufacturer’s instructions. ThedPCR reaction solutions were prepared with 20 ng ofcDNA, the QS3D reaction mix, 0,9 μM of primer forwardand reverse and 0,25 μM of each FAM and VIC labeledprobes (Additional file 7). Digital PCR was performedaccording to Santos et al. (2017). Data analysis andmanagement were performed using QuantStudio™ 3DAnalysis Suite™ software (https://apps.lifetechnologies.com/quantstudio3d/). Confidence level was set to 95% anddesired precision to 10%, in the Poisson Plus algorithmversion 4.4.10. Further statistical analysis was performedwith Statistica software (Statsoft Inc., Tulsa, USA). Afterconfirming the non-normal distribution using Shapiro-Wilktest, the Wilcoxon-Mann-Whitney test (non-parametric)was applied to the dPCR data, evaluating the variation be-tween time points and between combinations.

Additional Files

Additional file 1: Unsuccessful grafts detected throughout the field trialin the heterografts (A) and autografts (B). Percentage of unsuccessfulgrafts detected until the 80DAG are represented in blue, the onesdetected only at the end of cycle are represented in red. The percentageis calculated in relation to the total number of grafts. The success graftsare represented in green. (DOCX 19 kb)

Additional file 2: Heatmap of transcripts involved in hormone signaling,pathway signaling, regulation of gene expression (TF) and cell wallregulation. The transcripts were found DE between time points (21DAGand 80DAG), specifically DE in both combinations. A total of 108 DEGs inTN21/110R (A) and a total of 90 DEG in TN112/110R (B) are shown in allfour libraries separately. The transcripts are organized by functionalannotation. The asterix highlights the combination in which thetranscripts are statistically DE (FDR < 0.05). (DOCX 624 kb)

Additional file 3: Correlation between the TFs analyzed by MACE-Seqand digital PCR. (DOCX 16 kb)

Additional file 4: MicroRNAs selected for qPCR analysis, their annotatedpotential targets, and respective functions. (XLSX 75 kb)

Additional file 5: List of DEGs between combinations and betweentime points. (XLSX 592 kb)

Additional file 6: Primer sequences of Vvi- microRNAs used forexpression quantification by qPCR. (DOCX 14 kb)

Additional file 7: Primers and TaqMan®-Probes sequences used forgene expression quantification by dPCR. (DOCX 15 kb)

AbbreviationsABA: Abscisic acid; DAG: Days after grafting; DEG: Differentially expressedgenes; dPCR: Digital PCR; IAA: Auxin; miRNA: Microrna; qPCR: QuantitativePCR; TF(s): Transcription factor(s); TN112: Touriga Nacional clone 112; TN112/110R: TN112 grafted in the rootstock 110R; TN21: Touriga Nacional clone 21;TN21/110R: TN21 grafted in the rootstock 110R

AcknowledgmentsThe authors thank Teresa Roque, José Valadas and João Carvalho fromPlansel Nursery for the acquisition and management of the plants, as well asthe monitoring and maintenance of plants in the nursery; Margarida Teixeirafor her collaboration in the field work and harvesting of plants. The authorsalso acknowledge Susana Araújo for her contribution in the revision of thefinal manuscript.

Availability of data and materialAll relevant information is in Additional files 1, 2, 3, 4, 5, 6, 7. All raw RNAsequencing data used in this research have been submitted to NCBISequence Read Archive (SRA) under the BioProject PRJNA517111. Thesubmission id is SUB5084389, with the SRR attribution of SRR8487047,SRR8487050, SRR8487049 to TN21/110R at 21 DAG triplicates; SRR8487041,SRR8487042, SRR8487043 to TN21/110R at 80 DAG triplicates; SRR8487045,SRR8487046, SRR8487048 to TN112/110R at 21 DAG triplicates, andSRR8487044, SRR8487051, SRR8487052 to TN112/110R at 80 DAG triplicates.This data is accessible at https://www.ncbi.nlm.nih.gov/sra/PRJNA517111.

Authors’ contributionsEED and JB contributed to the experimental design and field trialexperiments, CS did part of the digital PCR experiment. MA performed allthe experiments, data analysis and wrote the manuscript. PF contributed todesign, data analysis and manuscript modification. All authors helped in therevision of the manuscript. All authors read and approved the finalmanuscript.

FundingThe design of the study and collection of data was funded by NationalFunds through FCT - Foundation for Science and Technology under theProject PTDC/AGR-PRO/118081/2010. Analysis and interpretation of the dataas well as the writing of the manuscript was funded by the PhD grant PD/BD/113476/2015 of MA, within the Research unit GREEN-it “Bioresources forSustainability” (UID/Multi/04551/2013), within the scope of the PhD programPlants for Life (PD/00035/2013).

Ethics approval and consent to participateNot applicable.

Consent for publicationNot applicable.

Competing interestsThe authors declare that they have no competing interests.

Author details1Plant Cell Biotechnology Laboratory, Instituto de Tecnologia Química eBiológica António Xavier (Green-it Unit), Universidade Nova de Lisboa,Apartado 127, 2781-901 Oeiras, Portugal. 2Genetics and Genomics of PlantComplex Traits (PlantX) Laboratory, Instituto de Tecnologia Química eBiológica António Xavier (Green-it Unit), Universidade Nova de Lisboa,Apartado 127, 2781-901 Oeiras, Portugal. 3Instituto Nacional de InvestigaçãoAgrária e Veterinária (Biotechnology and Genetic Genetic Resources Unit)INIAV-Dois Portos, Quinta da Almoínha, 2565-191 Dois Portos, Portugal.4Departamento de Biologia Vegetal, Faculdade de Ciências da Universidadede Lisboa, Campo Grande, 1749-016 Lisboa, Portugal.


https://apps.lifetechnologies.com/quantstudio3d/https://apps.lifetechnologies.com/quantstudio3d/https://doi.org/10.1186/s12870-019-1967-8https://doi.org/10.1186/s12870-019-1967-8https://doi.org/10.1186/s12870-019-1967-8https://doi.org/10.1186/s12870-019-1967-8https://doi.org/10.1186/s12870-019-1967-8https://doi.org/10.1186/s12870-019-1967-8https://doi.org/10.1186/s12870-019-1967-8https://www.ncbi.nlm.nih.gov/sra/PRJNA517111

Received: 15 March 2019 Accepted: 8 August 2019

References1. Lee J-M, Kubota C, Tsao SJ, Bie Z, Echevarria PH, Morra L, et al. Current

status of vegetable grafting: diffusion, grafting techniques, automation.Sci Hortic (Amsterdam). 2010;127:93–105. https://doi.org/10.1016/j.scienta.2010.08.003.

2. Legros PJP. L’Invasion du vignoble par le Phylloxéra. Academie des scienceset lettres de montpellier 1993. http://www.ac-sciences-lettres-montpellier.fr/academie_edition/fichiers_conf/Legros1993.pdf. Accessed 11 Sept 2017.

3. Errea P, Garay L, Marı JA. Early detection of graft incompatibility in apricot (Prunus armeniaca ) using in vitro techniques. Physiol Plant. 2001;112:135–41.

4. Usenik V, Krška B, Vičan M, Štampar F. Early detection of graftincompatibility in apricot (Prunus armeniaca L.) using phenol analyses. SciHortic. 2006;109:332–8.

5. Ciobotari G, Brinza M, Morariu A, Gradinariu G. Graft incompatibilityinfluence on assimilating pigments and soluble sugars amount of somepear ( Pyrus sativa ) cultivars. Not Bot Horti Agrobot Cluj-Na. 2010;38:187–92.

6. Pina A, Errea P, Juel H. Graft union formation and cell-to-cellcommunication via plasmodesmata in compatible and incompatible stemunions of Prunus spp. Sci Hortic. 2012;143:144–50. https://doi.org/10.1016/j.scienta.2012.06.017.

7. Goldschmidt EE. Plant grafting: new mechanisms, evolutionary implications.Front Plant Sci. 2014;5:1–9. https://doi.org/10.3389/fpls.2014.00727.

8. Melnyk CW. Plant grafting: insights into tissue regeneration. Regeneration.2017;4:3–14. https://doi.org/10.1002/reg2.71.

9. Herrero J. Studies of compatible and incompatible graft combinations withspecial reference to hardy fruit trees. J Hortic Sci. 1951;26:186–237. https://doi.org/10.1080/00221589.1951.11513736.

10. Ermel FF, Catesson AM, Poëssel JL. Early histological diagnosis of apricot/peach x almond graft incompatibility: statistical analysis of data from 5-month-old grafts. Acta Hortic. 1995:497–504. https://doi.org/10.17660/ActaHortic.1995.384.79.

11. Fernández-García N, Carvajal M, Olmos E. Graft union formation in tomatoplants: peroxidase and catalase involvement. Ann Bot. 2004;93:53–60.

12. Güçlü SF, Koyuncu F. A method for prediction of graft incompatibility insweet cherry. Not Bot Horti Agrobot Cluj-Napoca. 2012;40:243–6.

13. Assunção M, Canas S, Cruz S, Brazão J, Zanol GC, Eiras-Dias JE. Graftcompatibility of Vitis spp.: the role of phenolic acids and flavanols. Sci Hortic(Amsterdam). 2016;207:140–5.

14. Assunção M, Pinheiro J, Cruz S, Brazão J, Queiroz J, Eiras Dias JE, et al. Gallicacid, sinapic acid and catechin as potential chemical markers of Vitis graftsuccess. Sci Hortic (Amsterdam). 2019;246:129–35. https://doi.org/10.1016/j.scienta.2018.10.056.

15. Musacchi S, Pagliuca G, Kindt M, Piretti MV, Sansavini S. Flavonoids asmarkers for pear-quince graft incompatibility. J Appl Bot. 2000;74:206–11.

16. Cookson SJ, Clemente Moreno MJ, Hevin C, Nyamba Mendome LZ, Delrot S,Trossat-Magnin C, et al. Graft union formation in grapevine inducestranscriptional changes related to cell wall modification, wounding,hormone signalling, and secondary metabolism. J Exp Bot. 2013;64:2997–3008. https://doi.org/10.1093/jxb/ert144.

17. Cookson SJ, Clemente Moreno MJ, Hevin C, Nyamba Mendome LZ, Delrot S,Magnin N, et al. Heterografting with nonself rootstocks induces genesinvolved in stress responses at the graft interface when compared withautografted controls. J Exp Bot. 2014;65:2473–81.

18. Irisarri P, Binczycki P, Errea P, Martens HJ, Pina A. Oxidative stress associatedwith rootstock-scion interactions in pear/quince combinations during earlystages of graft development. J Plant Physiol. 2014;176:25–35. https://doi.org/10.1016/j.jplph.2014.10.015.

19. Melnyk CW, Schuster C, Leyser O, Meyerowitz EM. A developmentalframework for graft formation and vascular reconnection inArabidopsis thaliana. Curr Biol. 2015;25:1306–18. https://doi.org/10.1016/j.cub.2015.03.032.

20. He W, Wang Y, Chen Q, Wang X. Dissection of the Mechanism forCompatible and Incompatible Graft Combinations of Citrus grandis ( L .)Osbeck (‘ Hongmian Miyou ’). Int J Mol Sci. 2018;19(2):505. https://doi.org/10.3390/ijms19020505.

21. Renault-Spilmont A-S S, Boursiquot G and J-M. SYRAH DECLINE Report onthe meeting of the National Working Group – 11 April 2005. ProgrèsAgricole et Viticole. 122:337–348.

22. Cosgrove DJ. Loosening of plant cell walls by expansins. Nature. 2000;407:321–6.23. Abiri R, Shaharuddin NA, Maziah M, Yusof ZNB, Atabaki N, Sahebi M, et al.

Role of ethylene and the APETALA 2/ethylene response factor superfamilyin rice under various abiotic and biotic stress conditions. Environ Exp Bot.2017;134:33–44.

24. Chen C. Potentiation of developmentally regulated plant defense responseby atwrky18, a pathogen-induced Arabidopsis transcription factor. PlantPhysiol. 2002;129:706–16. https://doi.org/10.1104/pp.001057.

25. Chang X, Heene E, Qiao F, Nick P. The Phytoalexin resveratrol regulates theinitiation of hypersensitive cell death in Vitis cell. PLoS One. 2011;6:e26405.https://doi.org/10.1371/journal.pone.0026405.

26. Grimplet J, Van Hemert J, Carbonell-Bejerano P, Díaz-Riquelme J, DickersonJ, Fennell A, et al. Comparative analysis of grapevine whole-genome genepredictions, functional annotation, categorization and integration of thepredicted gene sequences. BMC Res Notes. 2012;5(213). https://doi.org/10.1186/1756-0500-5-213.

27. Söderman E, Hjellström M, Fahleson J, Engström P. The HD-zip gene ATHB6in Arabidopsis is expressed in developing leaves, roots and carpels and up-regulated by water deficit conditions. Plant Mol Biol. 1999;40:1073–83.

28. Yordanov YS, Regan S, Busov V. Members of the lateral organboundaries domain transcription factor family are involved in theregulation of secondary growth in Populus. Plant Cell Online. 2010;22:3662–77. https://doi.org/10.1105/tpc.110.078634.

29. Guerriero G, Sergeant K, Hausman J-F. Wood biosynthesis and typologies: amolecular rhapsody. Tree Physiol. 2014;34:839–55. https://doi.org/10.1093/treephys/tpu031.

30. Hirakawa Y, Kondo Y, Fukuda H. Establishment and maintenance of vascularcell communities through local signaling. Curr Opin Plant Biol. 2011;14:17–23. https://doi.org/10.1016/j.pbi.2010.09.011.

31. Suer S, Agusti J, Sanchez P, Schwarz M, Greb T. WOX4 imparts auxinresponsiveness to cambium cells in Arabidopsis. Plant Cell Online. 2011;23:3247–59. https://doi.org/10.1105/tpc.111.087874.

32. Pekker I. Auxin response factors mediate Arabidopsis organ asymmetry viamodulation of KANADI activity. Plant Cell Online. 2005;17:2899–910.

33. Hunter C. Trans-acting siRNA-mediated repression of ETTIN and ARF4regulates heteroblasty in Arabidopsis. Development. 2006;133:2973–81.

34. Korasick DA, Westfall CS, Lee SG, Nanao MH, Dumas R, Hagen G, et al.Molecular basis for auxin response factor protein interaction and the controlof auxin response repression. Proc Natl Acad Sci. 2014;111:5427–32. https://doi.org/10.1073/pnas.1400074111.

35. Rinaldi MA, Liu J, Enders TA, Bartel B, Strader LC. A gain-of-functionmutation in IAA16 confers reduced responses to auxin and abscisic acidand impedes plant growth and fertility. Plant Mol Biol. 2012;79:359–73.https://doi.org/10.1007/s11103-012-9917-y.

36. Perianez-Rodriguez J, Manzano C, Moreno-Risueno MA. Post-embryonicorganogenesis and plant regeneration from tissues: two sides of the samecoin? Front Plant Sci. 2014;5:1–11. https://doi.org/10.3389/fpls.2014.00219.

37. Etchells JP, Provost CM, Turner SR. Plant vascular cell division is maintainedby an interaction between PXY and ethylene Signalling. PLoS Genet. 2012;8:e1002997. https://doi.org/10.1371/journal.pgen.1002997.

38. Vahala J, Felten J, Love J, Gorzsás A, Gerber L, Lamminmäki A, et al.A genome-wide screen for ethylene-induced ethylene responsefactors (ERFs) in hybrid aspen stem identifies ERF genes that modifystem growth and wood properties. New Phytol. 2013;200:511–22.https://doi.org/10.1111/nph.12386.

39. Hussey SG, Mizrachi E, Creux NM, Myburg AA. Navigating the transcriptionalroadmap regulating plant secondary cell wall deposition. Front Plant Sci.2013;4(325). https://doi.org/10.3389/fpls.2013.00325.

40. Schuetz M, Smith R, Ellis B. Xylem tissue specification, patterning, anddifferentiation mechanisms. J Exp Bot. 2013;64:11–31. https://doi.org/10.1093/jxb/ers287.

41. Guo C, Xu Y, Shi M, Lai Y, Wu X, Wang H, et al. Repression of miR156 bymiR159 regulates the timing of the juvenile-to-adult transition inArabidopsis. Plant Cell. 2017. https://doi.org/10.1105/tpc.16.00975.

42. Wang Y, Kollmann R. Vascular differentiation in the graft union of in-vitrografts with different compatibility. Structural and Functional Aspects. J PlantPhysiol. 1996;147:521–33.

43. Yin H, Yan B, Sun J, Jia P, Zhang Z, Yan X, et al. Graft-union development: adelicate process that involves cell–cell communication between scion andstock for local auxin accumulation. J Exp Bot. 2012;63:4219–32. https://doi.org/10.1093/jxb/ers109.


44. Nanda AK, Melnyk CW. The role of plant hormones during grafting. J PlantRes. 2018;131:49–58. https://doi.org/10.1007/s10265-017-0994-5.

45. Dubrovina AS, Kiselev KV. Regulation of stilbene biosynthesis in plants.Planta. 2017;246:597–623.

46. Bavaresco L, Petegolli D, Cantù E, Fregoni M, Chiusa G, Trevisan M.Elicitation and accumulation of stilbene phytoalexins in grapevine berriesinfected by Botrytis cinerea. Vitis. 1997;36:77–83.

47. Chung IM, Park MR, Rehman S, Yun SJ. Tissue specific and inducibleexpression of resveratrol synthase gene in peanut plants. Mol Cell.2001;12:353–9.

48. Harju AM, Venalainen M, Laakso T, Saranpaa P. Wounding response in xylemof scots pine seedlings shows wide genetic variation and connection withthe constitutive defence of heartwood. Tree Physiol. 2009;29:19–25.

49. Li J. The WRKY70 transcription factor: A node of convergence forJasmonate-mediated and salicylate-mediated signals in plant defense. PlantCell Online. 2004;16:319–31. https://doi.org/10.1105/tpc.016980.

50. Pandey SP, Somssich IE. The role of WRKY transcription factors in plant immunity.Plant Physiol. 2009;150:1648–55. https://doi.org/10.1104/pp.109.138990.

51. Bakshi M, Oelmüller R. WRKY transcription factors: Jack of many trades inplants. Plant Signal Behav. 2014;9:1–18. https://doi.org/10.4161/psb.27700.

52. Cookson SJ, Ollat N. Grafting with rootstocks induces extensivetranscriptional re-programming in the shoot apical meristem of grapevine.BMC Plant Biol. 2013;13:147.

53. Weyman PD. A circadian rhythm-regulated tomato gene is induced byarachidonic acid and Phythophthora infestans infection. Plant Physiol. 2005;140:235–48. https://doi.org/10.1104/pp.105.068874.

54. Chen Z, Zhao J, Hu F, Qin Y, Wang X, Hu G. Transcriptome changesbetween compatible and incompatible graft combination of Litchi chinensisby digital gene expression profile; 2017. p. 1–12.

55. Espen L, Cocucci M, Sacchi GA. Differentiation and functional connection ofvascular elements in compatible and incompatible pear/quince internodemicrografts. Tree Physiol. 2005;25:1419–25. https://doi.org/10.1093/treephys/25.11.1419.

56. Pina A, Cookson S, Calatayud A, Trinchera A, Errea P. Physiological andmolecular mechanisms underlying graft compatibility. In: Colla G, Pérez-Alfocea F, Schwarz D, editors. Vegetable Grafting: Principles and Practices.Wallingford Oxfordshire: CABI; 2017. p. 132–54.

57. Fan M, Xu C, Xu K, Hu Y. Lateral organ boundaries domain transcriptionfactors direct callus formation in Arabidopsis regeneration. Cell Res.2012;22:1169–80.

58. Brackmann K, Greb T. Long- and short-distance signaling in the regulationof lateral plant growth. Physiol Plant. 2014;151:134–41.

59. Ji J, Strable J, Shimizu R, Koenig D, Sinha N, Scanlon MJ. WOX4 promotesprocambial development. Plant Physiol. 2010;152:1346–56. https://doi.org/10.1104/pp.109.149641.

60. Himmelbach A. Homeodomain protein ATHB6 is a target of the proteinphosphatase ABI1 and regulates hormone responses in Arabidopsis. EMBOJ. 2002;21:3029–38. https://doi.org/10.1093/emboj/cdf316.

61. Melnyk CW, Gabel A, Hardcastle TJ, Robinson S, Miyashima S. Transcriptomedynamics at Arabidopsis graft junctions reveal an intertissue recognitionmechanism that activates vascular regeneration. Proc Natl Acad Sci U S A.2018;115(10):2447–56. https://doi.org/10.1073/pnas.1718263115.

62. Chang S, Puryear J, Cairney J. A simple and efficient method for isolatingRNA from pine trees. Plant Mol Biol Report. 1993;11:113–6.

63. Wang L, Feng Z, Wang X, Wang X, Zhang X. DEGseq: an R package foridentifying differentially expressed genes from RNA-seq data.Bioinformatics. 2009;26:136–8.

64. Usadel B, Nagel A, Steinhauser D, Gibon Y, Bläsing OE, Redestig H, et al.PageMan: an interactive ontology tool to generate, display, and annotateoverview graphs for profiling experiments. BMC Bioinf. 2006;8:1–8.

65. Galli V, Guzman F, De Oliveira LFV, Loss-Morais G, Körbes AP, Silva SDA,et al. Identifying microRNAs and transcript targets in Jatropha seeds.PLoS One. 2014;9.

66. Ramakers C, Ruijter JM, Lekanne Deprez RH, Moorman AFM. Assumption-free analysis of quantitative real-time polymerase chain reaction (PCR) data.Neurosci Lett. 2003;339:62–6.

67. Wang C, Wang X, Kibet NK, Song C, Zhang C, Li X, et al. Deep sequencingof grapevine flower and berry short RNA library for discovery of novelmicroRNAs and validation of precise sequences of grapevine microRNAsdeposited in miRBase. Physiol Plant. 2011;143:64–81. https://doi.org/10.1111/j.1399-3054.2011.01481.x.

68. Andersen CL, Jensen JL, Ørntoft TF. Normalization of real-time quantitativereverse transcription-PCR data: A model-based variance estimationapproach to identify genes suited for normalization, applied to bladder andcolon cancer data sets. Cancer Res. 2004;64:5245–50.

69. Formey D, Sallet E, Lelandais-Brière C, Ben C, Bustos-Sanmamed P, Niebel A,et al. The small RNA diversity from Medicago truncatula roots under bioticinteractions evidences the environmental plasticity of the miRNAome.Genome Biol. 2014;15:1–21.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.


AbstractBackgroundResultsConclusion

BackgroundResultsAnalysis of grafting successGenes differentially expressed between the two heterografts at 21DAGGenes differentially expressed between the two heterografts at 80DAGTranscriptome profiling between 21DAG and 80DAGDifferential expression of transcription factors involved in the regulation of vascular differentiationExpression of post-transcriptional regulators at the graft interface

DiscussionTN21 and TN112 have different levels of graft compatibilityCompatibility seems to be related with the activation of signaling in an early graft stageGenes encoding oxidative stress and wound healing proteins are more expressed in the less compatible heterograft at 80DAGA higher expression of TF involved in cambium maintenance and vascular differentiation seems to be important in driving graft success at 80DAG

ConclusionMethodsPlant materialGrafting procedureRNA extraction and purificationmRNA sequencing and expression profilingmicroRNA sequencingqPCRDigital PCR

Additional FilesAbbreviationsAcknowledgmentsAvailability of data and materialAuthors’ contributionsFundingEthics approval and consent to participateConsent for publicationCompeting interestsAuthor detailsReferencesPublisher’s Note

Documents

Understanding the molecular mechanisms underlying graft ... · RESEARCH ARTICLE Open Access Understanding the molecular mechanisms underlying graft success in grapevine M. Assunção1*,